> I'm not sure why LLMNR uses only one multicast address.
> I see serious problems with that. A LLMNR query would
> wake up every dormant host, in this case. Why this may
> be a serious problem?

There was actually a reasonably thorough discussion of this issue during
the design of LLMNR. You can find it in the list archive. At one time,
there was a design proposal to use multiple multicast groups. The
group-id would be derived from the queried name. The proposal was
eventually discarded.

Part of the reason was implementation complexity. A hash based solution
requires to write the name in canonical form before hashing. Canonical
functions are hard to get right, especially with international names,
and could cause significant interoperability issues. 

Another reason was the general overhead of multiple multicast groups.
Many station interfaces can only handle a limited number of groups, and
fall back to promiscuous mode and software filtering after that number.
In switched environments, switches have to maintain routing state for
the multiple groups. The multiple groups may reduce the overhead of
individual queries, but there is a significant management overhead.

Given that, it was decided to just keep it simple, and use exactly one
group.

-- Christian Huitema

--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to